GRA 19703
Master Thesis
Thesis Master of Science
A Study on Employee Well-Being: Investigating the effects of implementing Artificial Intelligence in Human Resource Management practices.
Navn: Ingvild Faye-Schjøll, Jenny Christina Høye
Start: 15.01.2020 09.00
Finish: 01.09.2020 12.00
BI Norwegian Business School – Master Thesis
A Study on Employee Well-Being: Investigating the effects of implementing Artificial Intelligence in Human Resource
Management practices.
Hand-in date:
01.09/2020 Campus:
BI Oslo
Examination code and name:
GRA19703 Supervisor:
Anders Dysvik Programme:
Master of Science in Leadership and Organizational Psychology
This thesis is a part of the MSc programme at BI Norwegian Business School. The school takes no responsibility for the methods used, results found, and conclusions drawn.
Page i
Acknowledgements
The present thesis has been written as the final assignment of the Master programme in Leadership and Organizational Psychology at BI Norwegian
Business School in Oslo. The past two years have been filled with many academic endeavours and thanks to highly competent professors we now stand at the end of our long educational journey, ready to apply our knowledge into practice. We are excited for what the future will bring. It is inevitable that Covid-19 has affected the everyday life of everyone, nonetheless brought some implications to this study. Though, the support we have received from a number of people has kept us motivated throughout the process. Upon completion of this project we, therefore, wish to dedicate this section to show our gratitude to a number of people.
First and foremost, we would like to thank our supervisor Anders Dysvik for always being available for guidance throughout this process. BI Business School has been closed during an extended period of time, like many other corporations, putting a strain on face-to-face interaction. The importance of digital adoption through video-meetings due to home-office solutions has hence been essential to keep dialogue during our research. Nevertheless, we are grateful for his valuable and immense knowledge which he has shared with us, both in person and through his research. This has inspired the topic of our paper. It is also important for us to thank Winningtemp for collaborating with us on our study. They have welcomed us into their offices and allowed us insight into their inspiring and extraordinary company culture. With this, it should be mentioned that the thesis would not have been possible to execute without the unconditional support Jacob Österberg has given us over the past two years. To him we want to emphasise our deepest gratitude as he has gone beyond expectations in securing that the research could be conducted, even during challenging times. We thank you for always setting aside time for us through countless emails and video conferences. Last but not least, we would like to thank the six organizations and corresponding respondents contributing to our research and helping us secure data. Every contribution has been more valuable and appreciated than you will ever know.
Thank you all for your support.
Page ii Abstract
The following study investigates the growing digitalization in a contemporary society and, particularly, how it is influencing the human resource management field. With burgeoning technology, we look to new solutions on how managers can measure and uphold employee well-being by introducing artificial intelligence as a means. The innovative company, Winningtemp, is set to be the benchmark of our study as they have provided organizations with new solutions, allowing managers continuous insight into their employees’ work-life-cycle. The
emergence of digitalization over the past decades and the continuous evolvement of technologies justifies the relevance of our study. Nevertheless, the critical phase organizations are currently facing due to Covid-19, requiring rapid change and digital solutions in order to keep operating.
Human resource scholars emphasise how employee well-being has not been a central research topic within the field (Ho & Kuvaas, 2019). Arguably though, employee well-being and performance are interconnected, supporting the fact that employees’ health should be a topic of interest for managers. Kuvaas, Buch and Dysvik (2016) further emphasise how performance management should indeed be a continuous process, yet many organizations still rely on traditional methods, limiting performance management practices to once or twice per year. The present research, therefore, seeks to understand whether agile performance management tools might be a more sufficient solution in enhancing healthy and essentially prosperous workplaces.
A quantitative study of 93 individuals was conducted to compare companies utilising human resource tools assisted by artificial intelligence, with traditional tools. Specifically, the study found that by adopting agile tools to performance management practices, organizations generated higher levels of employee well- being in terms of social-psychological functioning. Further, the results show that employees exposed to new-technological tools report lower on economic leader- member exchange relationships. Nonetheless, managers and line-managers within these companies engage in human resource practices on a more frequent basis and are generally more satisfied with their tools to approach employees.
Page i Table of Content
1.0 INTRODUCTION... 1
2.0 THEORETICAL BACKGROUND AND HYPOTHESES ... 4
2.1 Human Resource Management ... 4
2.2 Well-Being and Stress ... 5
2.2.1 Well-Being ... 5
2.2.2 Stress ... 7
2.2.3 Self-Determination Theory ... 9
2.3 Performance Management ... 10
2.4 The Digital Transformation of HRM ... 12
2.4.1 Artificial Intelligence ... 13
2.5 Leadership and HR Tools ... 14
2.5.1 Enabling HR ... 14
2.6 Leader-Member Exchange Theory ... 16
2.6.1 Social and Economic Leader-Member Exchange ... 16
3.0 METHODOLOGY ... 19
3.1 Research Design ... 20
3.2 Sampling ... 20
3.3 Data Collection... 21
3.3.1 Participants ... 21
3.3.2 Measures ... 21
3.4 Ethical Considerations ... 25
4.0 ANALYSIS ... 26
5.0 RESULTS ... 28
5.1 Exploratory Factor analysis ... 28
5.2 Hierarchical Moderated Regression Analysis ... 33
5.3 Independent Samples T-test ... 35
6.0 DISCUSSION ... 37
6.1 Limitations and Practical Implications ... 39
6.2 Covid-19 and Future Research ... 41
7.0 CONCLUSION ... 42
8.0 REFERENCE LIST ... 43
APPENDIX ... 51
Page 1
1.0 INTRODUCTION
Over the past twenty years organizations have been exposed to significant
changes in the environments in which they operate. Chiefly as a result of the shift from a manufacturing and production economy into a service sector asset-based economy, namely the new economy (Shipton, Sparrow, Budhwar, & Brown, 2017). Labour in the new economy can be characterized by high knowledge intensity, web-based organizing, internet connections and the integration of work design, technology and services (Porter, 1998). Today, organizations worldwide may be facing a new leap of change as a direct consequence of Covid-19. The year of 2020 has exposed businesses to unanticipated transformations, forcing organizations to adapt to new work-solutions in order to survive. It is inevitable that individual jobs are becoming even more digital than ever, as a direct consequence of the global crisis. Managers hence have an essential role both in the creation of opportunities and the protection of employees’ health (Bernin and Theorell, 2003).
Iacovini (1993) defines changes as a “natural process – one that is carefully planned and orchestrated to move organizations from one capacity to another” (p.
65). Given the global uncertainty; enhancing, renewing and revitalizing existing work systems is vital for organizational survival (Shipton et al., 2017). The motivation behind change from an organization's perspective can thus generally be associated with the ambition to progress and exceed one’s current position.
Constantly fluctuating environments pose a threat to organizations if not handled right, as they need to adapt to changes more rapidly than before, in order to keep up with competition and the market in which it exists. The natural response during a major change is to hold on to the safe and familiar aspects of an otherwise confusing myriad of unknowns (Iacovini, 1993). Often, organizational inertia arises as a consequence of resisting change and failing to keep pace with the market (Jones, 2013, p. 300).
Change management, therefore, holds a central role in every organization with the intent to avoid resistance and to promote openness to change. One can assume that
Page 2
change management plays an especially important role in organizations that are dependent on technological advances, as innovation is to a large extent associated with growth (Tidd & Bessant, 2018). Particularly, this may be more salient now than ever, and organizations exhibiting innovation, rather than remaining committed to what worked in the past, are more likely to thrive. Yet, innovation presents challenges for organizations as moving away from accepted ways of working into new domains is not straightforward, especially given people's commitment to what worked well in the past (Shipton et al., 2017). That being said, innovation is not purely about high technology, but also consists of several other aspects such as identifying or creating opportunities, offering new ways of serving existing markets, growing new markets, rethinking services, meeting social needs and last but not least, improving operations (Tidd & Bessant, 2018, p.
4-5). Together, these components of promoting innovation can provide any organization with a competitive advantage. Boxall and Purcell’s (2003) thus points out organizational flexibility as an important organizational goal (as cited in Boselie 2014). This includes mental flexibility, which represents employees’
ability and willingness to change without resistance, essentially linked to agility, reflecting the organization's capability of adapting easily to changes in its external environment (Chanda et al., 2007).
Nevertheless, in environments where many organizations have access to the same technology, people make the real difference (Mondy & Martocchio, 2015). When managing change and understanding the power of individual employees, much depends on the organizational culture. Paradoxically, changing or modifying a corporate culture means that an otherwise consistent and enduring culture has to adapt to its environment (Trompenaars & Prud’Homme, 2004, p.171). It is important to find the balance between what cultural factors to retain and which to implement as new. There are a multitude of evaluations that change managers and leaders in general need to consider when trying to run a change strategy as
smoothly as possible. In order to accomplish a successful change, it is essential to be able to change people's attitudes, behaviour and cognition, through flexibility and good people management (Boselie, 2014). The changes currently taking place all around the world, affecting the everyday work-life of organizational members
Page 3
are challenging organizations to create opportunities for well-being and efficacy.
Especially, in order to minimize risks and find solutions to keep operating while at the same time securing employees' health.
The topic of human resource management (HRM) is hence more important than ever. In order to stay competitive, organizations must emphasise their human resources (HR) rather than traditional forms of competitive advantage (Ho &
Kuvaas, 2019). Still, Ho & Kuvaas (2019) clarifies how there is a wide acceptance that employee well-being does not receive equivalent attention from HRM
scholars. They further state that if we want to accurately quantify the benefits of HRM practices, we need to move beyond focusing on financial outcomes and increase the focus towards employee well-being. In 2014 the Swedish
organization Winningtemp (WT) introduced a revolutionary system assisting managers with the measurement and improvement of organizational members well-being. Assisted by artificial intelligence (AI), WT provides a HR-technology system making it easier for managers to focus on employees’ goals, challenges, and achievements. In turn, through continuous feedback sessions in addition to the annual performance appraisals, both managers and employees are given the
opportunity to take responsibility and focus on development that creates results.
Based on more than 600 scientific studies, the AI-based platform helps managers and leaders visualise employees’ development in real-time and act on insights that strengthen engagement and improve performance. Ultimately, the system aims to create thriving workplaces, encouraging managers to focus on the entire lifecycle of the employee, from onboarding to offboarding, as job satisfaction and
profitability are interconnected. This is realized through a system measuring the following nine indicators: leadership; job satisfaction; meaningfulness; autonomy;
work situation; participation; personal development; team-spirit; and commitment.
Organizational and individual development go hand in hand; hence the system provides continuous insight into how each individual feels and contributes along the way. Finally, recommended actions are provided to assist managers in leveraging the potential of their employees and essentially maximize the organization's bottom line (Winningtemp, 2020).
Page 4
HRM is a broad field of study and we will for the purpose of this paper narrow the scope down to HR practices concerning performance management. Accordingly, the aim of this research is to study the effects of adopting HR-tools assisted by AI, with the purpose of increasing the quality of HR practices. This leads us to the following research question:
Will the implementation of Artificial Intelligence enhance managers utilisation of Performance Management practices, and enable employees’ well-being?
2.0 THEORETICAL BACKGROUND AND HYPOTHESES 2.1 Human Resource Management
HRM involves management decisions aimed at achieving individual,
organizational, and societal goals (Boselie, 2014). Paauwe (2004) elaborates how the emphasis of HRM is the exchange relationship between employers and their employees. The shaping of this employment relationship transpires in a
continuous tension between added and moral values, whereas the former
represents the organization's economic side, the latter views employees as human beings with feelings, emotions, opinions, norms and values. This illustrates how employees are more than just resources for creating organizational success, and consequently, HRM practices and policies are fundamental in any organization consisting of human beings (Boxall, Purcell and Wright, 2007). Schuler and Jackson (1987) explain how HRM practices are systems developed with the intent to attract, select, develop, motivate and retain employees, as well as induce
organizational flourishment.
The mutual gain proposition builds on the notion that “what is good for the employee is also good for the employer, and the other way around” (Boselie, 2014, p. 106). Certainly, HRM is proven to be positively associated with outcomes such as satisfaction, as well as employee-employer relationship outcomes, which both in turn positively affect organizational performance and attitudinal outcomes. However, evidence also illustrates how this may not always
Page 5
be applicable, specifically, when the employer's best-interest does not align with the employee’s. The critical perspective, therefore, builds on the assumption that HRM is negatively associated with health indicators such as stress and strain, resulting in an increased probability of burnout tendencies (Van de Voorde, Paauwe & Van Veldhoven, 2012).
Beer, Spector, Lawrence, Quinn Mills, and Walton (1984) argues that employee well-being should be considered when evaluating the long-term consequences of any HRM system. Nevertheless, Legge (1989) suggests that HRM may result in morally problematic issues in situations where workers are being exploited. In spite of this, employee well-being has not been a central research agenda within the HRM field, as scholars have tended to focus on the link between HRM and performance instead (Boselie, Dietz, & Boon, 2005). Consequently, research examining the relationship between work stress and well-being has started to flourish (Ganster & Rosen, 2013).
2.2 Well-Being and Stress 2.2.1 Well-Being
Well-being is a complex construct that concerns optimal experience and
functioning (Ryan & Deci, 2001). Broadly defined, well-being “refers to people’s evaluations of their lives” (Diener, Suh, Lucas, & Smith, 1999, p. 213), or
similarly, “all the things that are important to how we think about and experience our lives” (Rath & Harter, 2010, p. 137). Narrowing it down, employee well-being is suggested to be restricted to job satisfaction. Hence, today’s most accepted and comprehensive conceptualization is suggested by Grant Christianson and Price (2007), who defines employee well-being as “the overall quality of an employee’s experience and functioning at work” (p. 52). Accordingly, employee well-being is operationalized as the presence of dispositional positive affect and the absence of dispositional negative affect (Diener, 1994). Current research on well-being has been derived from two general perspectives: the hedonic approach and the eudaimonic approach. Whereas the former focuses on happiness, viewing well- being in terms of pleasure attainment and pain avoidance, the latter focuses on
Page 6
meaning and self-realization, distinguishing well-being in terms of the degree to which a person is fully functioning (Ryan & Deci, 2001). In organizational science, the hedonic approach has been frequently studied in connection to job satisfaction and organizational commitment, while the eudaimonic approach in terms of meaning and engagement (Grant et al., 2007).
Moreover, scholars often concentrate on psychological and subjective well-being.
Psychological well-being consists of the three dimensions: psychological, physical and social (Ho & Kuvaas, 2019). Firstly, high psychological well-being occurs if a person “experiences frequent positive emotions such as joy and happiness, and infrequent negative emotions such as sadness and anger” (Bakker
& Oerlemans, 2011, p. 179). Secondly, the physical dimension is concerned with physiological indicators and subjective experiences of bodily health. In
organization science, this has often been studied in terms of injuries, diseases, job- related anxiety, stress, burnout and exhaustion (Grant et al., 2007). Finally, the social dimension addresses the quality of relationships at work, which has been widely studied in terms of trust, social support, reciprocity, leader-member exchange, cooperation, coordination, and integration (Grant et al., 2007). Good social relationships are traditionally defined as having the support of others, though recent work has emphasized that humans also need to support others.
Indeed, findings show how helping others can be more important to health than receiving help, as people gain more from giving to others than receiving from them (Diener et al., 2009).
Further, Diener (1984) suggests that subjective well-being (SWB) consists of three major components, including life satisfaction, positive experiences, and negative experiences. Evidence indicates that these components are independent and are influenced by different causes, yet, SWB has proved overall beneficial on health and longevity, social relationships, and productivity. Essentially, SWB is likely to exert a causal role in good health. Seligman (2002) argues that
individuals possess desirable feelings in addition to pleasant ones, these include, engagement or interest, and involvement in activities that are meaningful and purposeful. Indeed, there is found support that purpose and meaning are beneficial
Page 7
to human functioning and well-being (Seligman, 2002). Research explains how people who feel satisfied and experience positive feelings most of the time, act differently from those who are unhappy. Besides, studies illustrate that positive feelings can have desirable physiological effects, while negative experiences have undesirable effects (Diener, 2013). As living entities, we are born to flourish, meaning, within all individuals there is a natural propensity to actively engage, assimilate, and master one’s environments. Moreover, there is a desire to learn, grow, and where possible, to develop and express capacities, talents, and interests.
(Diener et al., 2009).
It is important to note, however, that there are universal differences in causes of SWB across the globe since the substance of having one's basic needs met and experiencing supportive social relationships varies between societies. For instance, there are found differences between individualistic and collectivistic cultures in regard to life satisfaction. On the one hand, people in individualistic societies, such as Norway and Sweden, pay more attention to their emotions when making life satisfaction judgments. This is due to the fact that personal emotions are considered to be a core component of individual identity. Hence, dimensions such as self-esteem and emotional feelings have a stronger correlation with life satisfaction in these cultures. On the contrary, people in collectivistic cultures pay more attention to their social relationships and whether others view their lives as successful. In other words, culture influences not only people’s levels of SWB but the factors that mainly affect it, meaning, what individuals consider to be most important concerning life satisfaction. Accordingly, people are generally happier and tend to have higher SWB if they have characteristics consistent with cultural norms (Diener, 2013).
2.2.2 Stress
In a constantly fluctuating world, where people are under continuous pressure to perform and to exceed expectations, stress has become a global condition, and a harmful one. Scholars struggle to form consensus on the conceptualization of stress because of its ambiguous and vague nature. For that purpose, numerous studies have been conducted, as well as a myriad of composed definitions in an
Page 8
attempt to encompass a precise formula of its complexity and multi-faceted origin.
This study will first and foremost concentrate on occupational stress, also known as work stress. Ganster and Rosen (2013) define work stress as the process by which workplace psychological experiences and demands produce both short-term and long-term changes in mental and physical health. Stress is embedded in an ongoing process that involves individuals transacting with their environments, making appraisals of those encounters, and attempting to cope with the issues that arise (Cooper, Dewe & O’Driscoll, 2001, p. 12).
Lazarus (1991) emphasizes that our well-being may be threatened when stress arises. In fact, stress makes people almost three times as likely to leave their jobs and can harm strategic thinking and creative abilities. Essentially, this can lead to burnout, which can generate absenteeism, turnover, declined productivity, in addition to medical, legal, and insurance costs. The consequences of this when unacknowledged, are detrimental for employee well-being and business
performance which could, in turn, threaten an organization's bottom line (Peart, 2019). This emphasises how leaders must commit to workplace wellness, in order to create healthier work environments (Peart, 2019). In fact, managers play a pivotal role by perpetuating employees’ work efforts in settings with high
competence-related stress (Dysvik, Kuvaas & Buch, 2014). A study conducted by Motowidlo, Packard and Manning propose that job-related stress often will lead to depression in employees which in turn impede their interpersonal and
cognitive/motivational performance (Motowidlo et al., 1986, p. 626). Another study suggests that stressors expedite cognitive fatigue and impair task
performance as a result of information overload and lower energy levels (Cohen, 1980).
The stress paradigm is vast, even when concentrated on occupational stress alone.
Thus, the focal point of this paper will be the postulation of stress as introduced by House (1979; 1980), which comprises the following elements of perceived job pressure; quality concern, responsibility pressure, workload and role conflict.
Firstly, House, Wells, Landerman, McMichael and Kaplan explain quality
concern as “having concern about not being able to do as good work as one could
Page 9
or should” (1979, p.141). Further, they interpret responsibility pressure as having
“too much responsibility for people, process, or products and insufficient human or material assistance” (House et al., 1979, p. 141). The third element, workload, involves reporting a large quantity of work and frequent time pressure (House et al., 1979, p.141). Hart and Staveland define workload as “the perceived
relationship between the amount of mental processing capability or resources and the amount required by the task” (1988). Besides, previous studies have addressed findings revealing high levels of occupational stress as a result of excessive overtime, amongst high-level managers and administrators as well as younger well-educated employees (Krantz, Berntsson & Lunberg, 2005). Finally, role conflict concerns receiving ambiguous and/or conflicting expectations from others at work (House et al., 1979, p.141). Another definition as proposed by House, Schuler and Levanovi, described in terms of incompatibility of expectations, implying that role conflict “occurs in response to the focal person’s perception of aversive (conflicting) stimuli (demands) in the environment” (1983, p. 337).
2.2.3 Self-Determination Theory
In today’s organizations, profitability alone is not enough as employee motivation and well-being are crucial contributors to long-term organizational success (Deci, Olafsen & Ryan, 2017). Self-determination theory (SDT) involves human
motivation and personality, and how people’s inherent growth tendencies and psychological needs are the foundation for self-motivation. The theory focuses on what facilitates high-quality, sustainable motivation, and also, what initiates engagement within employees (Ryan & Deci, 2000; Deci et al., 2017).
Specifically, the theory proposes that employees’ performance and well-being are affected by their level of motivation towards their job. With the focus on
differentiating approaches to motivation, SDT has identified various types of motivation, each with different consequences for learning, performance, and well- being (Ryan & Deci, 2000).
First, essential to SDT is the difference between autonomous motivation and controlled motivation. Whereas the former involves acting with a sense of choice, the latter entails engaging in activities because one has to (Gagné & Deci, 2005).
Page 10
SDT proposes that by fostering autonomy and support in the workplace, increased employee satisfaction and thriving may occur, in addition to collateral benefits for organizational effectiveness. Surely, more autonomous forms of motivation are associated with higher performance quality and employee well-being over time, as opposed to controlled motivation (Deci et al., 2017). Further, the theory
differentiates between intrinsic motivation; doing an activity because it is interesting and for the inherent satisfaction of the activity, and extrinsic
motivation; doing an activity to attain a consequence, such as tangible or verbal rewards (Gagné & Deci, 2005; Deci et al., 2017).
The impact of varied environment factors such as job design, pay contingencies, managerial styles on experiences and motivation are mediated by a set of basic psychological needs essential for all people, namely belongingness, competence, and autonomy (Deci et al., 2017). SDT research concerns the outcomes of the extent to which people are able to satisfy these needs in social environments (Gagné & Deci, 2005). Indeed, research implies that commitment and authenticity reflected in intrinsic motivation are more likely to be evident when people
experience support in these components (Ryan & Deci, 2000). In addition, researchers have found that workplaces that focus their attention towards satisfying the psychological needs of individuals, facilitate autonomous motivation, psychological and physical well-being, and essentially enhanced performance (Deci & Ryan 2000, in Deci et al., 2017).
2.3 Performance Management
Performance management (PM) is by DeNisi and Pritchard defined as “a broad set of activities aimed at improving employee performance” (2006, p. 255). The overall objective is to improve performance, productivity and effectiveness at the organizational level by motivating employees at the individual level. Moreover, the organization should aim to provide information that will assist managers in improving employee performance (DeNisi & Pritchard, 2006). Activities typical for PM include goal setting, evaluating goal attainment, and providing feedback (Kuvaas, Buch & Dysvik, 2016). These activities are often organized as formal meetings or as performance appraisals that are held once or twice per year.
Page 11
However, it is postulated that such evaluations, and PM in general, should be implemented as a continuous process (Kuvaas et al., 2016).
The effectiveness of PM is disputed in organizational theories. Among researchers that raise doubt about its effectiveness is Adler, Campion, Colquitt, Grubb,
Murphy, Ollander-Krane and Pulakos (2016), who suggest that methods such as performance ratings should be disregarded entirely. They propose that formality is what impedes traditional PM approaches. Furthermore, Pulakos and O’Leary (2011) supplement this argument by positing that the main issue of PM is that it has “been reduced to prescribed, often discrete steps within formal administrative systems that are disconnected from the day-to-day activities that determine performance management's effectiveness” (p. 146). This, in turn, causes the actuality and timing of the feedback to be problematic in that it lacks both flexibility and dynamism (Pulakos & O’Leary, 2011). Kuvaas and colleagues (2016) recognize that when performance-related evaluations such as goals, objectives, targets, key performance indicators, and performance standards, are reduced to once or twice a year, they are often inaccurate and obsolete, sometimes even plain wrong. Furthermore, these methods are often incompatible with the new and modern jobs, such as. virtual work, remote work, temporary work, semi- autonomous teams, freelancing and so forth, stemming from the continuously and ever-evolving nature of work (Pulakos and O’Leary, 2011).
PM, which at first was incorporated with the intention of motivating employees, has claimed to be counterproductive. In fact, most employees deem PM systems and performance appraisals as “frustrating, too bureaucratic, and often not
relevant to their job” (Adler et al., 2016, p. 221). Despite its adverse effects, PM is indeed very much utilized in organizations. Surely, one cannot ignore PM
completely, as performance, skills, attributes and assets inevitably will be judged or evaluated, either consciously or unconsciously by leaders or colleagues. Still, leaders and organizations need to look at new ways to achieve this. Amongst the various types of measurements of performance, it is crucial that the organization creates a starting point metric in which the populations share a common
Page 12
understanding of what is important for individual and collective performance (Adler et al., 2016, p. 234).
2.4 The Digital Transformation of HRM
In line with digital changes organizations have endured over the past decades, digital technologies play an increasingly prominent role for both employees and HRM (Parry, Strohmeier & Nickson, 2014). The relationship between technology and human interaction is considered authentic and necessary, as this alliance emerges as a means of supporting and enhancing strategic goals and objectives.
Hence, technology can be utilised as a tool of support and development of management and decision-making systems, essentially, contributing to the
remodelling of organizational practices (Jatoba, Gutierriz, Fernandes, Teixeira, &
Moscon, 2019).
Ruel and colleagues (2007) define e-HRM practices as “a way of implementing HR strategies, policies and practices in organizations through conscious and direct support of and/or with the full use of web-based technology channels” (as cited in Bissola & Imperatori, 2014, p. 378). This constitutes new management systems that enable relationship opportunities between employees and organizations. Yet, little attention has been devoted to relational e-HRM practices, which concerns systems specifically designed and implemented with the purpose of managing and sustaining relationships with employees (Bissola & Imperatori, 2014). Bissola and Imperatori (2014) found that new e-HRM practices can, in fact, influence the relationship between employees and HR departments. Moreover, the approach can be significant in sustaining employee trust towards the HR department, building more direct, individualized and regular relationships between employees and HR professionals.
Surely, cognitive technology has redesigned HR solutions, allowing HR professionals access and disseminate information more rapidly (Manuti & De Palma, 2017). A survey conducted by “Sungard Availability Services and Tame”
on British, American and European employees revealed that 81 percent of the employees consider having access to the latest digital tool as important.
Page 13
Moreover, 36 percent would leave their current employer for an organization more progressive to digital adoption. Nevertheless, digital transformation brought 42 percent greater staff retention, 56 percent improved workplace satisfaction and 60 percent increased productivity (Chakraborti, 2016).
2.4.1 Artificial Intelligence
According to Bellman (1978, in Jatoba et al., 2019), AI is the process of activity automation associated with human thinking activities such as decision making, problem-solving and learning. The technology enables faster and more efficient decision making, contributing to organizational development, by bringing people closer together. If applied properly, AI practices are assumed to make HRM more effective. Hence, AI is increasingly playing a significant role in HRM, for
instance within recruitment and employee development processes (Jatoba et al., 2019).
Nonetheless, AI may also be used to predict employee attrition, since employee turnover may be both costly and challenging. Here, the technology is used to accurately predict employee exit based on transactions generated by the employees and machine learning. This real-time data insight assists managers with employee satisfaction levels and future predictions of potential turnover. As to engagement, AI finds patterns and reasons that cause stress to employees which essentially affects their performance level. In this way, managers can identify potential issues and address them in a timely manner. Evidence shows how
employee experience is becoming a priority for employers to keep their workforce engaged, hence, AI may be used for virtual assistance, making it easier to meet the needs of every single employee (Bhatia, 2018).
Indeed, expectations on AI within the HRM field are sky-high, offering platforms to engage employees as co-creators of strong, smart and advanced workplace cultures. This progress has transformed the way employees perceive their jobs, work relationships with colleagues, and essentially their contribution to
organizational growth (Bhatia, 2018).
Page 14 2.5 Leadership and HR Tools
HR professionals should be competent in staffing, development, compensation and employee labour relations. Though, their competence goes beyond this, as it is also required that they have the ability to communicate and operate HR
practices to organizational members (Manuti & De Palma, 2017). Line managers (LMs) are becoming increasingly more involved in the implementation of HR practices as they play a vital role in bringing organizational policies and practices to life. In HRM, particular attention has been given to how LMs perceive and make sense of organizational policies, practices and strategies, in addition to how they influence the success or failure of the implementation of these practices.
However, research suggests that there may be dispute between employees, LMs’
and managers' perceptions of HR practices within organizations (Kuvaas, Dysvik
& Buch, 2014).Dysvik and Kuvaas (2012) have revealed a positive relationship between employees' perceptions of supportive or relational LMs, and employees' perception of HR practices. In other words, the relationship between LMs and HR is crucial for efficient implementation of practices (Kuvaas et al., 2014). Kuvaas and colleagues (2014) suggest that a possible approach for improving the
efficiency of HR implementation may be to make the process more motivating for LMs, by increasing the ease of implementation and making them more “user- friendly”.
2.5.1 Enabling HR
Kuvaas, Dysvik and Buch’s (2014) define enabling HR (EHR) as “LMs
perception of the extent to which their organization's HR practices assist them in their managerial responsibilities, and the degree to which LMs are provided with the discretion/autonomy and flexibility to take local and individual needs into account when implementing HR practices” (p. 848). Being provided with sufficient autonomy can hence be contemplated as a symbol of trust in LMs competence. This, in turn, may enhance a sense of responsibility for outcomes and willingness to work for organizational goals. Essentially this should increase LMs’ intrinsic motivation to take local and individual needs into account through
Page 15
a proper HR implementation and profoundly grasping the responsibility for people issues (Kuvaas et al., 2014).
Guest (2011) emphasises how failure to implement HR policies as intended is a well-documented challenge. It is recognized that the effectiveness of HR practices is indeed influenced by the extent to which supervisors adopt those practices.
Although most managers intend to undertake HR activities; limited time, unclear policies and procedures, as well as lack of competence and support may impede implementation (Nehles et al., 2006; Evans, 2015, as cited in Williams, 2019).
Clearly, LMs do not have the same knowledge as HR professionals, thus support and guidance from HR departments in regard to implementation and their people management responsibilities is necessary. This may comprise policies and procedures, information and advice, tools and training to develop knowledge and skills, and technology in regard to the quality of HR systems. The outcomes may be that supervisors perceive HR tasks to be less of a burden (Williams, 2019).
In essence, autonomy and discretion are not just critical components of perceived enabling HR, but also influence the overall evaluation of the extent to which LMs think that their organization’s HR practices assist them in their managerial
responsibilities. Surely, LMs are likely to play an essential role in employees' sensemaking of their work environment, hence the more they themselves perceive HR practices as enabling, the more they should be prosocially motivated to
support their employees through HR implementation (Kuvaas et al., 2014). As such, we formed the following hypotheses for our study:
Hypothesis 1: Leaders perception of enabling HR will be positively related to employee well-being, while at the same time negatively related to stress.
Hypothesis 2: Leaders' level of engagement in HR tools will affect well-being; a more frequent use will have a positive effect on well-being, while an infrequent
use will have a negative effect on well-being.
Page 16 2.6 Leader-Member Exchange Theory
The leader–member exchange (LMX) theory is a relationship-based, dyadic theory of leadership based on the assumptions that leaders influence their employees through the quality of the relationships they develop with them (Erdogan & Bauer, 2015). According to the theory, leaders develop unique exchange relationships of varying quality with individual followers (Kuvaas, Buch, Dysvik, & Haerem, 2012). This exchange relationship shapes the expected behaviours of both parties (Furst & Cable, 2008), and are assumed to fall on a range from low-quality transactional-based relationships to high-quality relationships (Kuvaas et al., 2012). Interactions in low LMX relationships are transactional and impersonal (Furst & Cable, 2008). In this relationship, both the leader and the follower expect direct reciprocity characterized by short-term economic exchanges of behaviours (Kuvaas et al., 2012). On the contrary, interactions in high-quality LMX relationships represent a social exchange relationship, characterized by loyalty, emotional support, mutual trust, liking and professional respect (Furst & Cable, 2008; Erdogan & Bauer, 2015). Additionally, in these relationships, leaders typically provide support, developmental
opportunities, mentoring, and other benefits to the employee. The advantage of this can be increased member motivation to reciprocate to the leader, through loyalty and higher levels of voluntary behaviours (Erdogan & Bauer, 2015). In spite of this, Sparrowe and Liden (1997) point out one particular challenge in relation to applying social exchange theory to LMX research. Specifically, they state how “the dimensions of actual exchange behaviour that differentiate economic from social exchange, have not been specified in a way that facilitates empirical verification” (p. 524).
2.6.1 Social and Economic Leader-Member Exchange
Kuvaas and colleagues (2012) argue that social leader-member exchange (SLMX) and economic leader-member exchange (ELMX) relationships represent different forms of relationships. Hence, they propose that instead, the exchanges are to be considered as two distinct constructs and not merely as one continuum (Kuvaas et al., 2012). Whereas, SLMX represents the quality of social and relational
Page 17
interactions where long-term orientation, trust and investment are essential aspects of the exchange, aligning with a high-quality LMX relationship, ELMX is
typically represented by exchange relationships where employees reciprocate to their leader based on formal requirements, rather than mutual respect and interest (Kuvaas et al., 2012). An ELMX relationship may, therefore, be considered as more short-term and impersonal, resting on formal status differences (Buch, Martinsen & Kuvaas, 2015). Buch (2012) asserts that this does not entail a lack of leadership though, but rather specific patterns of interaction among leaders and employees directed towards a more transactional and calculative exchange relationship. ELMX relationships can, in fact, encourage subordinates to engage in behaviours that meet, but not exceed, the organization's expectations (Shore, Tetrick, Lynch & Barksdale, 2006).
Moreover, Dulebohn, Bommer, Liden, Brouer and Ferris (2011) found that social relationships could indeed contain factors which are normally associated with a low-quality exchange relationship. Surely, economic or instrumental behaviours linked with a low-quality relationship can exist over time and remain as the relationship flourishes into a higher quality relationship (Goodwin, Bowler &
Whittington, 2009). This justifies how leadership styles may hold different qualities, rather than different levels of quality (Kuvaas et al., 2012). Judge and Piccolo (2004, as cited in Kuvaas et al., 2012) argue that essentially both ELMX and SLMX may motivate productive behaviours. Still, Kuvaas and colleagues (2012) research found that SLMX relationships were positively related to work performance and organizational citizenship behaviour within employees, while ELMX relationships, on the contrary, were negatively related to these outcomes.
Hence, there still seems to be some dispute in the existing literature on the topic of ELMX.
Nevertheless, prior studies have investigated how the relationship with one's immediate leader may be a core indicator in reducing subordinates' influence of stressors such as high job demands, and essentially increase aspects of
psychological well-being (Cohen, 2004). Specifically, supervisors may do this through social support, which is by Cohen (2004) defined as “a social network’s
Page 18
provision of psychological and material resources intended to benefit an
individual’s ability to cope with stress” (p. 676). This can be done by clarifying tasks, communicating goals and professional development opportunities, and by providing job resources and feedback (Northouse, 2007). Based on the notion that social support reduces occupational stressors and strain, while having an overall positive relationship with well-being (Cohen, 2004), we suggest the following hypotheses for our study:
Hypothesis 3: SLMX will be positively related to well-being, while negative related to stress.
Hypothesis 4: ELMX will be negatively related to well-being, while positively related to stress.
Organizations are essentially cooperative systems that rely on the willingness of members to behave in certain ways that support the organization (Furst & Cable, 2008). As individual goals often differ from those of the organization, a primary responsibility of managers should be to persuade members to direct their efforts towards the organization's goals. During organizational change, the importance of this may be particularly salient (Erdogan & Bauer, 2015). Certainly, empirical evidence shows how members' perceptions of the relational qualities towards their immediate leader are essential to their perception and responses towards HR practices (Kuvaas et al., 2012). Hence, we hypothesise that the relationship between leaders perceived enabling HR and employee well-being and stress, will be moderated by the degree of SLMX and ELMX relationship:
Hypothesis 5: The relationship between enabling HR and well-being will be moderated by a) SLMX and b) ELMX; a) the higher SLMX, the more positive the
relationship, while b) the higher ELMX, the more negative the relationship.
Hypothesis 6: The relationship between enabling HR and stress will be moderated by a) SLMX and b) ELMX; a) the higher SLMX, the more negative the relationship, while b) the higher ELMX, the more positive the relationship.
Page 19
In sum, the presented research shows that there is a clear gap in the literature in relation to HR-tools assisted by AI. To our knowledge, there are no previous studies examining the efficiency of using such tools, in regard to well-being and stress. Hence, we seek to extend the HRM literature by studying the effects of using AI-assisted HR-tools in HR practices, in comparison to traditional tools.
This leads us to the final hypothesis of our study and essentially our proposed model:
Hypotheses 7: The use of HR tools assisted by AI will have a more favourable effect on well-being and stress, than the use of Traditional HR tools.
Figure 1. Proposed model
3.0 METHODOLOGY
The following section will cover the research methodology used to conduct our study. Leedy and Ormrod (2001) define research methodology as “the general approach the researcher takes in carrying out the research project” (p. 14). In order to answer the initial research question we conducted a quantitative study which
“involves the utilization and analysis of numerical data using specific statistical techniques to answer questions like who, how much, what, where, when, how many, and how. It is also described as the methods of explaining an issue or phenomenon through gathering data in numerical form” (Apuke, 2017, p. 46).
Page 20 3.1 Research Design
A cross-sectional design is by many considered to be the preferred research design in quantitative studies. This method entails “collection of data on more than one case and at a single point in time, in order to collect a body of quantitative or quantifiable data in connection with two or more variables, which are then
examined to detect patterns of association” (Bell, Bryman & Harley, 2019, p. 53).
Moreover, the study follows a descriptive research approach within the
quantitative field, which involves the “identification of attributes of a particular phenomenon based on an observational basis, or the exploration of correlation between two or more phenomena” (Williams, 2007, p. 66). As quantitative studies encompass numbers and aggregated data, it is crucial for the accuracy of the results that the data collection and processing is done correctly.
3.2 Sampling
For the purpose of this study, we utilised a convenience-based sampling as we needed organizations using, specifically, WT services. The organizations included in our sample thus represent a wider range of companies using HR tools assisted by AI. The sample consisted of 48 individuals from three companies, where 13 (27,1 percent) of these had a leader responsibility. Moreover, among the total respondents, 19 (39,6 percent) were women and 29 (60,4 percent) were men, with an average age of 26-35 years (50 percent). A second sample was introduced in the study as a control group for the companies using HR tools assisted by AI. This sample was included in the study to increase the validity of our findings.
Specifically, it is composed of three companies using traditional HR tools, and comprised 45 individuals, where 22 (48,9 percent) of these had a leader
responsibility. Furthermore, the entire sample consisted of 30 (66,7 percent) women and 15 (33,3 percent) men, with also here an average age of 26-35 years (42,2 percent).
Page 21 3.3 Data Collection
To test our hypotheses and essentially the initial research question, two web-based questionnaires were designed through Qualtrics. Surely, within a descriptive research approach, surveys are the most common methodology for data collection (Williams, 2007). Hence, our data was collected through a standardized
questionnaire and was distributed electronically through email. To avoid any linguistic misinterpretations, the questionnaires were written and conducted in English. This to strengthen the construct validity.
3.3.1 Participants
The questionnaire designed for companies utilizing HR tools assisted by AI (WT) were distributed to 95 email addresses within three Swedish companies, in June 2020. The complete responses we collected after sending out two reminders consisted of 50,53 percent. To anonymize the participants and their corresponding company, we will be referring to them as ‘AI Companies’ in the following
analysis. The questionnaire to companies using traditional HR tools, on the other hand, were distributed in July 2020. This because the control group was included as an extension of our research to secure sufficient data. Here we distributed 155 emails to three Norwegian companies, where a total of 29 percent responded.
These companies will further on be referred to as ‘Traditional Companies’.
3.3.2 Measures
Leader-Member relationship
First, to measure the employees' perceived relationship to their immediate leader, we used the LMX-measure presented by Kuvaas and colleagues (2012). Here, participants were asked eight questions, where four were related to SLMX, and the respective four ELMX. The variables were measured on a 7-point Likert scale ranging from “strongly disagree” = 1, to “strongly agree” = 7. The questions concerning ELMX comprised: “The most accurate way to describe my
relationship with my immediate leader is that I do what I am told to do''; “I do
Page 22
what my immediate leader demands from me, mainly because he or she is my formal boss”; “My relationship with my immediate leader is mainly based on authority, he or she has the right to make decisions on my behalf and I do what I am told to do''; and “All I really expect from my immediate leader is that he or she fulfils his or hers role as a supervisor or boss” (Kuvaas et al., 2012, p. 763). While SLMX, on the other hand, include: “My relationship with my immediate leader is based on mutual trust”; “My leader has made a significant investment in me”; “I try to look out for the best interest of my immediate leader because I can rely on him or her to take care of me”; and “The things I do on the job today will benefit my standing with my immediate leader in the long run” (Kuvaas et al., 2012, p.
763).
Enabling HR
Leaders' perception of enabling HR (EHR) was measured through Kuvaas and colleagues (2014) measure of generic user-friendly HR-tools for leaders. On a 7- point Likert scale ranging from “strongly disagree” = 1 to “strongly agree” = 7, leaders were asked five questions concerning their company's HR tools. However, in the questionnaire distributed to companies using HR tools assisted by AI we rephrased the questions from generally including traditional HR tools and systems, to specifically covering WT (Appendix 2).
Well-being
We assessed the measure of subjective well-being using Diener and Biswas- Diener’s (2009) Scale of Positive and Negative Experience, further referred to as SPANE. The scale assessed positive (SPANE-P) and negative (SPANE-N) feelings on a 12-item questionnaire, with six positive and six negative measures.
Specifically, respondents were asked to think about what they had been doing and experiencing the past four weeks, before reporting on a 5-point Likert-scale ranging from “very rarely or never” = 1 to “very often or always” = 5. Because the scale includes general positive and negative feelings, it assesses the full range of positive and negative experiences, including any specific feelings that could be unique in particular cultures. Due to the general items included, the scale can
Page 23
assess not only the pleasant and unpleasant emotional feelings, but also reflect other states such as interest, flow, positive engagement, and physical pleasure (Diener et al., 2009).
Further, we measured social–psychological prosperity (SPP), to complement existing measures of subjective well-being, through Diener and Biswas-Diener (2009) Flourishing Scale. This scale was used since it includes content that goes beyond psychological well-being narrowly defined (Diener et al., 2009), capturing a more accurate picture of the measure. The scale is an eight-item scale
encountering social relationships; having supportive and rewarding relationships, contributing to the happiness of others, and being respected by others. In addition to having a purposeful and meaningful life and being engaged and interested in one’s activities. Items tapping into self-respect and optimism were also included.
Finally, the scale measured an item of feeling competent and capable in the activities that are important to the respondent. Hence, the brief scale assesses major aspects of social–psychological functioning from the respondent’s own point of view (Diener et al., 2009). Through a 7-point Likert scale, respondents were then asked to range their answers concerning the items from “strongly disagree” = 1, to “strongly agree” = 7.
Stress
It is important to note that one of the greatest challenges when measuring and studying the effects of stress upon performance, is to isolate stress from all other variables, as stress varies significantly amongst people and situations. Hence, it is almost impossible to measure stress perfectly due to its multidimensionality. A complete measurement of performance under stress on an individual level would require an assessment of each subject, their personality and abilities, as well as other influencing aspects (Lazarus, Deese & Osler, 1952). Due to limited time and resources, this extensive method will not be manageable for this study. Hence, stress was measured through the instrument of House (1980), though, for the purpose of this study, the perceived occupational stress measures have been adapted to fit the environment of white-collar workers. The respondents were thus asked four questions related to the four different measures of job pressure (House,
Page 24
1979; 1980): quality concern; “the amount of work you do interferes with how well the work gets done”, responsibility pressure; “you do not have enough help and resources to do the job well”, workload; “you do not have enough time to get the job done well”, and role conflict; “you have to try to satisfy too many different people”. Essentially the respondents reported on a 5-point Likert scale ranging from “never” = 1, to “always” = 5 their level of agreement.
Dependent variables
We included the two dependent variables to the study. First, HR-Engagement
“how often do you engage in your company's HR tools” was measured on a 5- point Likert scale ranging: daily, weekly, monthly, yearly, to never. This was to uncover how frequently leaders and LMs actually engage in HR tools. However, we realize in after hand that the timeline between “monthly” and “yearly” is quite broad. It could hence have been a more sufficient measure if we had captured the item “quarterly” as well. Further, we also wanted to address to which degree leaders and LMs are satisfied with their HR tools through the measure of “HR- Satisfaction”. This was encountered by asking the following question: “would you recommend your company's HR tools to other companies”. The respondents were also here asked to indicate their level of agreement on a 5-point Likert scale ranging: definitely yes, probably yes, may or may not, probably not, to definitely not.
Control variables
Control variables were included in the study to outsource alternative explanations for the observed relationships. Cogliser, Schriesheim, Scandura, & Gardner (2009) suggests how this can increase internal validity as they might confound results. Hence, our control variables included age, gender, tenure, working hours and leader-responsibility. Age and gender were included in the study to control for potential demographic variance, whereas gender was measured as a
dichotomous variable (male = 1, female = 2), age was measured by six grouped categories (25 or less = 1, 26-35 = 2, 36-45 = 3, 46-55 = 4, 56-65 = 5, 66 or more
= 6). Moreover, the participants’ tenure and working hours per week were also
Page 25
included as control variables. Buch, Kuvaas, Dysvik and Schyns (2013) suggests how followers with a longer period of employment might respond differently to ELMX and SLMX relationships. Thus, tenure was included and measured through five grouped categories ranging (less than one year = 1, 1-3 years = 2, 4-6 years = 3, 7-9 years = 4, ten years or more = 5). Working hours were also accounted for as it could be interesting to see, specifically, if this potentially had any correlation to, for instance, higher levels of stress or lower levels of well-being. The variable was measured through the following categories (less than the average = 1, average = 2, more than the average = 3). Additionally, the respondents working more than the average of 37,5 hours in Norway and 40 hours in Sweden, were asked how many additional hours per week they spend on work to account for eventual extreme values. Finally, we include leader-responsibility as a control variable due to its relationship with the variables enabling-HR, SLMX and ELMX. This variable was coded as a dichotomous variable (yes = 1, unsure =2, no = 3).
3.4 Ethical Considerations
Naturally, there are several ethical considerations that are important to address in regard to the study (Johannessen et al., 2016). First of all, we informed the participants about the purpose of the research project, to ensure that they were fully aware of how the collected data was to be used. Furthermore, it was important to stress that we treat all personal data with full confidentiality and anonymity in accordance with General Data Protection Regulation, as the research should not be detrimental to the participants. Also, by developing a consent form which participants had to fill out before initiating the study, we emphasised how the participation was voluntary, and how the participant had the right to withdraw from the study at any time. Finally, in accordance with the Norwegian Center of Data regulations, the participants were informed about the collected data being deleted once the thesis has been submitted, September the 1’st 2020.
Page 26 4.0 ANALYSIS
The data analysis stage fundamentally concerns reducing the large scope of information gathered, in order to make sense of it. The raw data was hence processed, coded and analysed in the statistical software program SPSS Statistics.
In accordance with a cross-sectional design, we examined the data collected across units, or in this circumstance companies, to detect patterns of association (Bell et al., 2019). The data was hence analysed in several stages.
First, we tested for multicollinearity, normality and homoscedasticity as we wanted to ensure that our data was not impaired by outliers, errors or other insufficiencies. The data did not indicate any issues of multicollinearity as the coefficient's highest VIF value was 4.28, while the lowest was 1.1 which is above the commonly used cut-off point of .10 (Hair et al., 2010). Further, we did not detect any missing values, however, we did observe how five individuals report that they are “unsure” as to whether they have a leadership responsibility or not.
Hence, we might be missing out on valuable data in regard to “enabling HR”.
Moreover, we detected several outliers, specifically in regard to “working overtime”. Some respondents have reported working 45, 50 and 60 hours more than average working hours in their respective country. This shows a clear gap from other responses and hence we can assume that the question has been misinterpreted as to perhaps how many hours in total they work during a week.
Hence, we chose to remove these values from the data set. Though four
respondents still report working 15, 19, 25 and 30 hours overtime per week, yet we chose to retain these values as one can assume that individuals with a leader responsibility, within industries such as finance and sales, do indeed work this much overtime. Remaining outliers in other variables were retained to avoid causing any potential interference and misrepresentation of our results.
Secondly, we performed an exploratory principal component analysis with
varimax rotation on all multiple-scale items in the measurement model, in order to determine item retention (Medsker, Williams, & Holahan, 1994). Besides, this analysis is arguably a well-suited approach for evaluating and increasing both convergent and discriminant validity (Hurley Scandura, Schriesheim, Brannick,
Page 27
Vandeberg & Williams, 1997). Moreover, given that our study relies on self- report measured, we applied a conservative rule of thumb in the analysis and only retained items equal to a loading of .40 or higher. This was done to avoid
confounded measures and to address the concern of discriminant validity (Hair, Black, Babin, Anderson, & Tatham, 2010). Further, we only retained items with cross-loadings of less than .35 between observed factors (Kiffin-Petersen, &
Cordery, 2003), and a differential of .20 or higher between the included factors (Van Dyne, Graham, & Dienesch, 1994). Lastly, in order to get more
comprehensive information about our data, we computed the descriptive analysis and correlations for the finalized data.
Third, in order to test the studies hypotheses, we performed a hierarchical moderated regression (Cohen West, Aiken, & Cohen, 2003). Before computing the interaction terms, however, we centred the variables SLMX, ELMX and EHR by subtracting the mean of each variable from their corresponding score (Aiken &
West, 1991). This was done since interaction terms often create multicollinearity problems as a result of their correlations with main effects (Kuvaas, Buch, &
Dysvik, 2012). According to Aiken & West (1991), this procedure reduces the potential for these issues, in addition to facilitating the interpretation of the interaction (as cited in Buch, 2012). We also created dummy variables for the following nominal variables: gender; age; tenure; working-hours; leader responsibility; HR satisfaction; and HR engagement in order to conduct the analysis. Furthermore, in the regression analysis we followed the recommended practice by Aiken and West (1991) and first entered the control variables in step one. Further, the independent variable EHR was included together with its dependent variables HR satisfaction and HR engagement in step two, before the moderation variables ELMX and SLMX were inserted in the third step. Finally, the interaction terms EHRxELMX and EHRxSLMX were added in step four. The variables in the hierarchical moderated regression model accounted for 36,5 percent of the variance in well-being (R² = .365) and 42.4 percent of the variance in stress (R² = .424), with a significance of (p < .001). Fourth, to test the final hypothesis of our study, we conducted an independent t-test on our two samples:
Page 28
AI companies and Traditional companies, in order to test if there was a significant difference between the two samples' mean scores.
5.0 RESULTS 5.1 Exploratory Factor analysis
Based on the exploratory factor analysis, nine of our items did not fit the inclusion criteria of .40. Three items of stress were removed, namely “The amount of work you do interferes with how well the work gets done”; “You do not have enough help and resources to do the job well”; and “You have to try to satisfy too many different people”. Further, neither of the following items of SPP met the inclusion criteria either: “I am engaged and interested in my daily activities”; “I actively contribute to the happiness and well-being of others”; “I am a good person and live a good life”; “people respect me”. Nevertheless, the item “contended” and
“pleasant” from the measure of positive experiences was removed for the same reason. Further, the three SLMX items: “my immediate leader has made a significant investment in me”; “I try to look out for the best interest of my immediate leader because I can rely on him or her to take care of me”; and “the things I do on the job today will benefit my standing with my immediate leader in the long run” had cross-loadings above .35, in addition to a differential of less than .20 between its target construct and one of the other constructs. Also, the SPP item “my social relationships are supportive and rewarding” had a cross-loading of more than .35. These items were thus removed in order to increase convergent and discriminant validity. The rotated component matrix (Appendix 1) shows an overview of factor loadings above .40 after rotation, which could be considered significant (Field, 2009).
In order to determine discriminant and convergent validity, we deliberate on the results from our descriptive statistics reported in Table 1, together with the factor loadings from our analysis, with what we expected to find based on theory (Buch, Kuvaas, & Dysvik, 2011). Firstly, in accordance with Buch and colleagues (2011) findings, we expected a negative correlation between ELMX and SLMX, as these measures should be theoretically distinct from each other. However, we did not
Page 29
find any significance to this (-.042, p > .05) as shown in Table 1. In extended studies, Kuvaas and colleagues (2012) found that ELMX was significantly predicted by gender. Specifically, that male subordinates are more inclined to develop ELMX relationships than female subordinates. Hence, we sought to find whether there was any truth to this in our study, as our two samples were skewed in regard to gender representation, possibly affecting our results. Still, our
findings show no significance in relation to this (.211, p > .05). Interestingly though, there was instead a significance between ELMX and age (-.322, p < .01), as the younger generation reports higher levels of a transactional relationship with their immediate leader. It is reasonable to assume that younger employees, newer to the labour-market may view their superiors in a sense of formality, as one may get more assertive with age.
Additionally, Kuvaas and colleagues (2014) found that the more LMs’ perceive HR tools as enabling, the more open they are to implement them. Accordingly, we assume a stronger perception of enabling HR should be positively related to how frequently LMs’ utilise HR tools, in addition to how satisfied they are with the corresponding tools. Our factor analysis does indeed show that these items are related, falling under the same factor. Yet, as we see no significance in the relationship EHR and HR engagement (.264, p > .05), we can only confirm a significant correlation between EHR and HR satisfaction (.456, p < .01).
Nonetheless, we can also establish a positive correlation between HR tool satisfaction and engagement (.750, p < .01). Moreover, previous research shows that the more autonomy LMs are given, the more likely they are to implement HR tools (Kuvaas et al., 2014). Thus, we assume that SLMX should be positively related to both enabling HR and the frequency of HR tool utilization. Based on our results we can indeed establish a positive relationship between SLMX and perceived enabling HR (.451, p < .01). Furthermore, our factor analysis shows that the SLMX item “my relationship with my immediate leader is based on mutual trust” falls into the same continuum as HR engagement, yet there was not enough evidence for any significance in the relationship (.147, > .05). On the contrary, our findings do suggest a significant negative relationship between
